pub fn schema_fingerprint(columns: &[SchemaColumn]) -> StringExpand description
Compute the canonical schema fingerprint for a set of columns.
The fingerprint is "xxh3:<16-char-lowercase-hex>". The algorithm prefix
is part of the format so future hashers (sha256, blake3) can coexist —
readers MUST verify the prefix before interpreting the hex body.
Canonicalization: columns are sorted by name (case-sensitive) and
serialized as <name>\t<data_type>\n joined with no separator. Column
order in the source schema does not affect the fingerprint, but
column names and types do; renaming or retyping a column changes
the fingerprint.
This is the value written to the manifest’s schema_fingerprint field
(ADR-0012 M3) and is what --validate compares against to detect
schema drift between the time of write and the time of verify.